Michael's voice memo today sets the next build: an app for Signature's agents, bigger than the business planner, built around five pillars he is sending as a blueprint. The line that matters most in the memo is the picture of the product: you open it and it tells you what to do today, which transactions are moving, which client you have not spoken to in six days, and that the offer you submitted yesterday needs a chase. This page works that memo into three product shapes, an idea list per pillar, and a plan for the agent session next Wednesday, 30 September. The blueprint slots into it when it lands.
Build A, The Morning Page, as the front of the app. Take C's numbers as a strip inside it and use B's voice capture as the way work gets in, not as the interface. Serhant shipped a chief of staff yesterday, so the Signature edge has to be the day plan wired to Signature's own data and rules, not another chat box.
Take the five pillars to the agents on Wednesday as questions, with the sketch below on the screen, and let them tell us which pillar they would open the app for. Nothing here is built.
Tour Dash was the trial and it landed, so the next project is the one Michael has wanted to build for Signature: an agent operating system. The reference is Serhant's S.MPLE, which he rates as weak once you know what AI can do now. The bar is a system that is more useful day to day than that.
Imagine you have an app or web version, you just open it up and it's telling you exactly, hey, this is what you need to do today. These are your transactions, these are the clients you're working with. You haven't spoken to this person in six days. You submitted an offer yesterday.Voice memo, 24 September 2026
Michael said "many other things as well". Transactions and a broker view are the two this page adds, because the memo's own example leans on both.
Worth knowing before Wednesday, because the agents may bring it up and because the timing changed this week.
A request desk for Serhant agents only. You send a voice note, email or in-app request and a mix of AI and a human "advisor" team returns the work: CMAs, listing presentations, just-listed posts, email blasts, flyers, market reports, follow-up scheduling, GCI tracking.
It is a concierge with a queue. The agent still has to know what to ask for.
Yesterday at their Sell It conference Serhant announced a 2.0 with an assistant called Dot that reads the agent's calendar, email, CRM and transactions and starts work without being asked, using what they describe as seventy-plus specialised agents, with Fair Housing and RESPA checks built in. Ryan Serhant's line: "chatbots give you answers, S.MPLE 2.0 gives you outcomes."
Rollout begins 5 October across their markets in Q4.
The proactive day plan is no longer a differentiator on its own. Serhant will ship one, at least on paper. Signature's edge is narrower and real: it is Signature's data, Signature's ninety-day track and Academy, New Jersey's attorney-review deals, and a brokerage small enough to wire the app to how its agents actually work, which a 2,000-agent rollout cannot do.
It also cannot be judged yet. Nobody outside Serhant has used 2.0.
The app opens as one page that already knows your day.
Every morning the app writes the page from four sources: transactions and their dates, contacts and the last touch on each, yesterday's logged activity, and the agent's place on the ninety-day track. Each line is a job with a reason under it and one action on the right. Do the action and the line clears. The strip at the bottom is the day's numbers against the plan from the roadmap calculator.
Everything else, the CRM, the marketing drawer, the transaction files, sits behind this page and is reached from a line on it.
The agent who does not want a system, only to be told what to do next. New agents on the track, because the page carries their onboarding items in the same list as their deals. Leadership, because the same engine produces a broker page of who is behind.
The page is only as good as the data behind it. If the CRM is a mess and nobody logs calls, the page is empty or wrong on day one, so pillar III (clean-up) has to ship before or with it. The reasons under each line must be true and specific, or agents will stop trusting the page within a week.
You talk, it does. The closest shape to what Serhant announced.
One conversation, voice first. The agent narrates the day between showings and the assistant turns each message into records: a contact note, a task, a draft, a change to a deal. It comes back with questions when it needs a decision. Work it produces waits in a queue for a tap to send.
Agents who live in the car and will never type into a CRM. Capturing the things that today get lost between the showing and the office.
This is a me-too of Dot, launched by a brokerage with a far bigger budget, and Michael's own point in the memo is that people who only ask a chat box questions are underusing AI. A chat is also blank when you open it. It does not tell you what to do, it waits. Keep the voice capture. Do not make it the front door.
Numbers first. The business planner grown up.
Starts from the goal-to-daily-number calculator already on the roadmap and makes it the whole app. Agents log activity in a two-minute evening close-out, the app keeps the pace, and leadership sees the same board across the office. Marketing and CRM live in tabs behind it.
Accountability. Signature's ops team, who already run the ninety-day track this way. Agents who respond to a number.
A scoreboard nags. If logging is the only thing the app asks of an agent, the app becomes homework and the numbers go stale. It also does none of the work; the memo wants a system that does things, not one that counts them.
Each card has a few things the app could do inside that pillar and one line on where Signature can beat a concierge desk. These are starting points for Wednesday, to be cut down by what the agents say.
Ready before you ask.
Nobody goes quiet by accident.
Ten contacts a day, not a weekend.
Told, not tracked.
The list is made. Go call it.
The same engine, read from the top.
| A · The Morning Page | B · The Chief of Staff | C · The Scoreboard | |
|---|---|---|---|
| Opens to | Today's list, with reasons | A blank thread | Goal, pace, numbers |
| Agent's job | Do the next line | Tell it what happened | Log the day |
| Does work for the agent | Yes, queued from events | Yes, on request | No |
| Depends on | Clean CRM and transaction dates | Good speech capture, good judgement | Agents logging honestly |
| Closest to | Serhant's stated 2.0 | Serhant's Dot | The business planner today |
| Signature's edge | Its own data, track and market | None on its own | Already runs this way in ops |
| Wednesday reaction | Show the sketch, ask what is wrong on it | Ask who would actually talk to their phone | Ask who would log every evening |
| First pillar to ship | III, then IV | III via voice | IV |
Michael wants a few agents in the room, different ones from last time. Run it as a listening session with one thing on the screen, not a demo.
Before Wednesday this sketch can be made into a clickable page with the room's agents' names on it, in the Evening tokens, so the reaction is to something real.
Build A. The memo describes it almost line for line, and it is the shape that does work for the agent instead of waiting to be asked. The list with a reason under every line is the whole product; the CRM, marketing drawer and transaction files are where the lines lead.
Ship pillar III first, quietly. Nothing on the morning page is true until the contacts are clean and the deals have dates. Six weeks of ten-a-day triage plus a voice note capture gets an agent's data to the point where the page can be trusted, and it is the one pillar the competition skips.
Treat Serhant's announcement as the deadline, not the threat. They will tell 2,000 agents they have a chief of staff from 5 October. Signature can put a working morning page in front of thirty agents who helped design it. That is the story to be able to tell by the end of the year.